MIT researchers will lead a project worth up to $53 million, funded by the U.S. Advanced Research Projects Agency for Health (ARPA‑H) under its STREAM program, to use generative artificial intelligence to create safer, more effective herbicide alternatives. Over one billion pounds of herbicides are applied in the United States each year; they protect crops and enable regenerative farming, yet many weeds have evolved resistance and exposure is linked to cancer, metabolic and endocrine disorders, affecting farm workers and the general public. The effort, called Precision Learning for On‑target Weed control (PLOW), is headed by Christopher Voigt, head of MIT’s Department of Biological Engineering, and includes partners from the MIT‑Harvard Broad Institute, University of North Carolina‑Chapel Hill, University of Tennessee‑Knoxville, University of Illinois‑Urbana‑Champaign and Robigo. The team is organized into three groups aligned with STREAM technical areas. The generative‑AI group will build models that predict molecules—either small chemicals or siRNA—that disrupt the same plant enzymes targeted by glyphosate, such as EPSP synthase, while optimizing synthesis cost, resistance risk, and safety for humans and ecosystems. This group is led by MIT chemical engineer Connor Coley, MIT AI‑and‑Health professor Regina Barzilay, and UNC pharmacy professor Alexander Tropsha. The models are expected to generate billions of candidates, which will be screened by a high‑throughput testing group headed by MIT biological engineer Angela Koehler and Broad Institute’s Jaime Cheah. Screening results will be fed back to improve the computational models. After screening, a third group will evaluate the shortlisted molecules in greenhouse trials; members include Neal Stewart and Scott Lenaghan of UT‑Knoxville’s Center for Agricultural Synthetic Biology, Patrick Tranel of the University of Illinois, Tim Willson of UNC’s Eshelman School of Pharmacy, and Robigo founder Andee Wallace. The first greenhouse tests of RNAi candidates are slated for completion by the end of year one. In parallel, the project will develop living microbial systems—engineered bacteria, fungi, and srRNA—that degrade agrochemicals before they reach food, soil, or waterways, converting them into biomass and providing a scalable way to cut environmental and human exposure. Funding is provided by ARPA‑H under agreement 140D042690030; the views expressed are those of the authors alone.
Review: By marrying cutting‑edge AI with agricultural chemistry and synthetic biology, the project tackles herbicide resistance, human health concerns, and environmental contamination in a single pipeline, offering a promising route to more sustainable food production and reduced public‑health risk.